Indoor Occupancy Monitoring Users Guide

Table of Contents

Introduction

This example demonstrates the use of Ti mmWave radar sensors in indoor settings and the ability to locate and track humans in confined spaces, including large rooms. This example goes over how to use a modified configuration to tune the out of box demo for the IWRL6432 device, and demonstrates how it can track both stationary and moving humans in a room within a 4 meter x 10 meter room.

Requirements

🛑 Before Continuing!
Please make sure you are able to run the Out of Box Demo for IWRL6432 before proceeding forward.

Hardware Requirements

Item Version/Requirement
mmWave Radar Sensor IWRL6432BOOST
Tripod At least 2 meters in height
Micro-USB/Power Cable
Computer PC with Windows 10

Software Requirements

Item Version
mmWave SDK 5.1.x.x
UniFlash 8.1.1+
MATLAB RUNTIME 9.11

Setup

The following steps show how to set up the hardware in order to perform indoor occupancy detection:

Hardware Setup

  1. Install IWRL6432BOOST in the middle of the room.
  2. Raise sensor to height of 2 meters
  3. Angle sensor forward to an angle of 15 degrees

Please refer to the images below for a reference setup

Quickstart

  1. Open UniFlash

  2. Set the device to flashing mode and power cycle the device

  3. Flash the device with the prebuilt binary for presence and motion detection from the SDK

    • Follow the instructions for UniFlash
    • The prebuilt binary can be found at <SDK_INSTALL>\examples\mmw_demo\presence_motion_detect\prebuilt_binaries\
    • Use the Release version of the appimage when flashing

  4. Switch the device to functional mode and power cycle the device

  5. Navigate to the visualizer found at <RADAR_TOOLBOX>\tools\visualizers\Low_Power_Visualizer\ and run the lowpower_demo_visualizer_6432.exe executable

  6. In the pop-up window, select the correct user COM port

    • Find the user COM port in the device manager

    • For the data port, select the same COM port as the user port

  7. Browse for a configuration, and select the indoor_occupancy_monitoring.cfg from <RADAR_TOOLBOX>source\ti\examples\Indoor_Occupancy_Monitoring\chirp_configs

  8. Click on Load Configuration and then press Done

Understanding the Output

The following videos below show the various scenarios in which this example can be used to localize humans in an indoor setting. The scenarios show detection for humans who are stationary in a cluttered environment as well as detection for humans who are moving in a cluttered environment. This configuration uses minor motion to enable detection even when no major movement is present in the scene, though some performance is slightly sacrificed when observing the point-cloud during movement.

Stationary Monitoring

The video below shows the ability to track stationary humans while they are standing up or sitting down. The point-clouds stay in the same place as where the human in the video, and updates this point-cloud as it detects minor motions like fidgeting and slight movement in the stationary positions. With this, the visualizer is able to indicate whether there is a person in the room, even if they are not moving. The size of the room and what constitutes a detection can be modified in the configuration based on the environment.

Dynamic Monitoring

The video below shows the ability to track moving humans in a crowd. As the humans move closer together, the point clouds can get mixed up due to a lag while using minor motion. However, if there is a reasonable amount of space between people, you can distinguish between stationary targets and moving targets.

Configuration Statistics

Statistic Value
Maximum Range 10.55 m
Range Resolution .3662 m
Maximum Velocity 4.08 m/s
Velocity Resolution 8.1522 m/s
Average Power 4.217 mW

Developer’s Guide

Modifying Configuration

This configuration has been tuned to improve the detection of humans in an indoor setting. In this scenario, the maximum range the radar can detect is 10 meters. This resulted in setting a boundary within this 10 meters. However, this can be modified based on the various environments that may exist in any room. For example, some rooms may require a much larger boundary box horizontally than vertically, while other rooms might be much smaller. Adjustments to the boundary box and sensor position can be modified via the mpdBoundaryBox and sensorPosition commands. In addition, in some rooms, you may end up getting less reflections and as a result smaller point clouds, while other rooms may cause extra(“ghost”) reflections and much more fluctuating point clouds. This can be tuned in two different ways: the CFAR threshold and the minorStateCfg.

Need More Help